zhangshenghang opened a new pull request, #13121:
URL: https://github.com/apache/seatunnel/pull/13121
Yes, I am contributing to Apache SeaTunnel.
## What documentation issues were found
While cross-checking connector documentation against the Java source code
and comparing English (`docs/en`) with Chinese (`docs/zh`) pages, I found three
groups of real, verifiable issues:
1. **Wrong "column projection" feature claims** — the docs' Key features
matrix claimed column projection support for connectors whose source classes do
not implement `SupportColumnProjection`
(`org.apache.seatunnel.api.source.SupportColumnProjection`):
- `source/Airtable.md`, `source/Jira.md`, `source/OneSignal.md`,
`source/Shopify.md` (all extend the HTTP base source, which does not implement
the interface)
- `source/Kudu.md` (`KuduSource` does not implement the interface)
2. **Inverted claim for the file source family** — `source/OssJindoFile.md`
unchecked column projection, although the shared `BaseFileSource` implements
`SupportColumnProjection` and every other file source doc (LocalFile, HdfsFile,
OssFile, ObsFile, BosFile, CosFile, GcsFile, S3File, SftpFile, SmbFile,
FtpFile) marks it as supported.
3. **English/Chinese option tables diverged in structure**:
- `docs/en/connectors/source/OssJindoFile.md`: the options table had only
4 columns (no `description` column), while the Chinese page has full 5-column
tables — added the English `description` column (37 options).
- `docs/zh/connectors/sink/Hudi.md`: both tables (base configuration +
table list) lacked the `描述` column — added it (24 options).
- `docs/zh/connectors/sink/Hive.md`: the options table lacked the `描述`
column — added it (18 options).
4. **Deprecated/unsupported env key in examples** — example configs used
`execution.parallelism` in `env {}` blocks. This key is a deprecated Flink-only
alias (`ConfigKeyName.PARALLELISM`, logs a deprecation warning) and is silently
ignored by the Zeta engine, which only reads `parallelism`
(`EnvCommonOptions.PARALLELISM`). Replaced it with `parallelism` in pages not
touched by other open PRs: `OceanBase-CDC`, `Opengauss-CDC`, `PostgreSQL-CDC`,
`Kafka` (source), `Intercom`, `Asana`, `transforms/metadata` (en + zh).
## What was changed
- Feature-matrix checkbox corrections (`[x]` ↔ `[ ]`) for column projection,
verified against `SupportColumnProjection` implementations in the code.
- Added missing description columns to option tables, with descriptions
aligned with the sibling-language page and the code (`FileBaseSourceOptions`,
`HudiSinkOptions`, Hive sink options). Hudi option defaults (`batch_size=1000`,
`batch_interval_ms=1000`, `insert/upsert_shuffle_parallelism=2`,
`min/max_commits_to_keep=20/30`, `index_type=BLOOM`, `record_byte_size=1024`,
`cdc_enabled=false`) were re-verified against `HudiSinkOptions` before
translating.
- Replaced `execution.parallelism` with `parallelism` inside `env {}`
example blocks.
## Areas/files updated (28 files, en + zh always in pairs)
- `docs/{en,zh}/connectors/source/`: Airtable, Jira, OneSignal, Shopify,
Kudu, OssJindoFile (feature claims; OssJindoFile en also got the description
column)
- `docs/{en,zh}/connectors/source/`: OceanBase-CDC, Opengauss-CDC,
PostgreSQL-CDC, Kafka, Intercom, Asana (env key fix)
- `docs/{en,zh}/transforms/metadata.md` (env key fix)
- `docs/zh/connectors/sink/Hudi.md`, `docs/zh/connectors/sink/Hive.md` (描述
column added)
## Both languages checked
Yes — every fix was applied to both the English and Chinese page. The en/zh
option-table row counts now match for the fixed pages, and all `plugin_name`
examples, option tables, and feature claims of both locales were checked.
## Duplicate PR check (last 7 days)
Checked all PRs created since 2026-10-27 (`gh pr list --state all`),
including today's #13117, #13118, #13119, #13113 and the open docs PRs #13050,
#13072, #13097:
- #13050 / #13072 / #13097 cover
Clickhouse/Doris/Mysql/Oracle/PostgreSql/Redis/SqlServer/StarRocks/TDengine/Databend/Elasticsearch/Hbase/Maxcompute/ObsFile/FakeSource/Paimon/Pulsar/Redshift/RocketMQ/SelectDB/Snowflake/MySQL-CDC/TiDB-CDC
and format docs — none of my 28 files overlap.
- #13118 (Split transform) — no overlap.
- #13119 (zh architecture pages), #13117 (changelog table), #13113 (Mem0) —
no overlap.
None of the open/merged PRs in the last week touches the feature-matrix
claims, the OssJindoFile/Hudi/Hive tables, or the `execution.parallelism`
examples listed above.
## Verification
- Automated checks written for this change (run locally over the whole
`docs/` tree):
- Relative-link check: 0 broken links after the change.
- Heading-anchor check: 0 new broken anchors (2 pre-existing ones in
`rest-api-v1.md` are left untouched because PR #13097 already covers that file).
- Markdown table integrity: all rows/separators in the 28 changed files
have consistent column counts.
- en/zh option-table row counts now match for OssJindoFile, Hudi, Hive.
- Full docusaurus website build (the CI `document` job) was not run locally
because it requires checking out the separate `apache/seatunnel-website`
repository and installing its npm dependencies; the changes are markdown-only
with consistent tables, inline code spans around all `${...}` placeholders, and
valid relative links/anchors, so the MDX render risk is minimal.
__zcode_status=$?
if [ "$__zcode_status" -eq 0 ]; then pwd -P >
'/var/folders/67/72cc9wwd6cl2fzm7rg17xxrh0000gn/T/zcode-631e7b6b-7da2-4e8d-9861-f04b0cb47626-cwd';
fi
exit "$__zcode_status"
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
